Proc. Natl. Acad. Sci. USA Vol. 84, pp. 8169-8173, November 1987 Neurobiology Blood- barrier protein recognized by monoclonal antibody (immunocytochemistry/peroxidase-antiperoxidase/Langerhans cells/experimental allergic encephalomyelitis/retina) NANCY H. STERNBERGER AND LUDWIG A. STERNBERGER Departments of Anatomy, Neurology and Pathology, University of Maryland School of Medicine, Baltimore, MD 21201 Communicated by Berta Scharrer, August 7, 1987

ABSTRACT An IgGi mouse monoclonal antibody pro- body is a useful probe for studying macromolecules related to duced in response to immunization with rat brain homogenate blood-brain barrier function. reacted with endothelial cells in the central and peripheral . Because antibody reactivity was associated MATERIALS AND METHODS with endothelia that have a selective permeability barrier, the antibody was called anti-endothelial-barrier antigen (anti- Monoclonal Antibody Production. Monoclonal antibodies EBA). Paraffin sections of Bouins'-fixed rat tissue were used were produced from BALB/c mice immunized with rat brain for initial screening and subsequent characterization of anti- homogenates as described (12) by using the procedure of body reactivity. The antibody was generally unreactive with Kohler and Milstein (13). Immunocytochemistry on paraffin endothelial cells in other organs and with nonendothelial cells sections was used to select hybridomas producing antibodies in or outside of the nervous system. Antibody binding was to endothelia. greatly reduced or absent in endothelia of the area postrema Immunocytochemistry. Lewis or Sprague-Dawley rats and choroid plexus, sites known to possess fenestrated blood were perfused with Bouins' fixative. Organs were postfixed vessels. In developing rat brain, anti-EBA binding to some in the same fixative overnight, dehydrated, and embedded in microvessels was seen at 3 days postnatally. Anti-EBA reac- paraffin. Seven-micrometer-thick sections of brain, periph- tivity outside the nervous system occurred in spleen and skin. eral nerves, optic nerve, retina, heart, lung, liver, intestines, Patchy reaction with portions of some spleen blood vessels and thymus, lymph nodes, spleen, adrenal, skeletal muscle, skin, binding to some cells in the spleen were observed. In the skin, pancreas, pituitary, and pineal body were treated with the small cells, tentatively identified as Langerhans cells, which antibody or reagent controls. Unfixed, frozen , Vibra- participate in Ia presentation, were stained. On immunoblots tome sections of Bouins'-fixed brains, and paraffin-embed- of rat brain microvessel preparations electrophoresed in Na- ded brains, fixed in either buffered 4% paraformaldehyde, or DodSO4/polyacrylamide gels, anti-EBA reacted with a protein buffered 2% paraformaldehyde with 2% glutaraldehyde, were triplet of Mr 30,000, 25,000, and 23,500 components. also examined for ability to bind the antibody and for pattern of reactivity. Among a number of monoclonal antibodies with selective The sections were treated in the following sequence: (i) 3% specificity for individual components of the central nervous normal sheep serum for 30 min; (ii) anti-EBA ascites fluid system that we have developed (1), one is an endothelial cell diluted 1:1000 to 1:500,000 for 1 hr to overnight; (iii) goat antibody (anti-endothelial-barrier antigen) (anti-EBA) that anti-mouse immunoglobulin, diluted 1:40 for 30 min; (iv) reacts with and peripheral nervous peroxidase-antiperoxidase complex (14, 15) prepared from system endothelia. Brain or nerve microvessel endothelial monoclonal antiperoxidase, diluted 1:100 to 1:800 for 30 min; cells are the site of the blood-brain or blood-nerve barrier and (v) 0.05% diaminobenzidine tetrahydrochloride/0.01% important in determining the selective permeability charac- hydrogen peroxide for 8 min. Wash solutions were 0.05 M teristic of the nervous system (2). Little is known about the Tris, pH 7.6, in 1.5% sodium chloride (1.5 T). Diluents were proteins associated with endothelia and endothelial-barrier the same solutions containing, in addition, normal sheep properties. Monoclonal antibodies can serve as useful probes serum. Micrographs were made with Nikon-Nomarski optics for elucidating the biochemical basis of the endothelial in order to visualize background. barrier. Microvessel Preparation. Brain microvessels were pre- A variety of interesting monoclonal antibodies (3-10) and pared from frozen rat brains by the methods described by an antiserum (11) have been reported that reacted with brain Mrsulja et al. (16) or by Lidinsky and Drewes (17). Two microvessels. Some of these antibodies reacted broadly with millimolar EGTA, aprotinin (0.5 trypsin inhibitor unit/ml), endothelia from other, non-nervous system organs (5-10). An leupeptin (10 ,g/ml), 2 mM phenylmethylsulfonyl fluoride, antiserum, prepared against isolated bovine brain microves- and 1 mM bacitracin were present during homogenization and sels, reacted with a Mr 46,000 protein found in brain but not in subsequent steps to prevent possible degradation of EBA in heart, liver, and kidney capillaries (11). Rat-brain micro- during the isolation procedure. Protein determination was vessels were used as the immunogen to produce a monoclo- done by the bicinchoninic acid assay (18). nal antibody that bound to brain endothelia, the brush border Endothelial Cell-Membrane Preparation. The microvessels of proximal kidney tubules, and bile canaliculi (3). A Mr were suspended in distilled water, agitated at 4°C, and 74,000 protein has been defined by a monoclonal antibody centrifuged at 15,000 x g for 10 min as described by Lidinsky that is expressed on the surface of chick embryonic blood and Drewes (17) to lyse endothelial cells and separate cells, brain endothelium, choroid plexus epithelium, and membranes from cytoplasmic materials. basolateral membranes of kidney tubules (4). Polyacrylamide Gel Electrophoresis. One-dimensional dis- In the present study, we describe a monoclonal antibody continuous gels were run in a vertical slab gel apparatus by directed against a triplet of endothelial proteins. This anti- the method of Laemmli (19) as has been described in detail previously (20). The stacking gel contained 4% acrylamide, The publication costs of this article were defrayed in part by page charge and the separating gel contained 8% acrylamide. Samples payment. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. §1734 solely to indicate this fact. Abbreviation: anti-EBA, antibody to endothelial-barrier antigen.

8169 Downloaded by guest on September 27, 2021 8170 Neurobiology: Sternberger and Sternberger Proc. NatL Acad. Sci. USA 84 (1987)

_w _ ,ft ArNW_ A> ft - _ * to 'gi 0 M .1a -*.o 9f4 S 9 , p 0 C, . AP:N . b41' ..,e s w U o 0 I, 0 0 4ow - At~~~~~11L .: 'r qr~m

lI " ilk* 4um 05~~~~~_#91 *a 0 t,,,,uf = ^0 C *~~~ U~~~~~~1 I . .. - 1 0ee . 0 FIG. 1. Sagittal section of adult rat brainstem treated with FIG. 3. Section treated first with anti-EBA, then counterstained anti-EBA. Antibody binding was limited to microvessels. (x120.) with hematoxylin/eosin to show association of antibody reactivity with endothelial cell. Arrows indicate endothelial nuclei. (x480.) were incubated for 3 min at 1000C in buffer containing 63 mM Tris HCI, pH 6.8/2% NaDodSO4/10% glycerol/5% 2-mer- All dilutions were made with 1.5x T buffer containing 1% captoethanol/0.001% pyronin Y and were used immediately. normal sheep serum. Molecular weight standards were run in adjacent lanes. Electrophoresis was done at constant voltage (150 V) for 4.5-5 hr with cooling. The gels were either stained in 0.125% RESULTS Coomassie blue R-250 in 50% methanol/10%o acetic acid and Anti-EBA reacted with microvessels in the adult rat brain destained in 5% methanol/7% acetic acid or were used (Fig. 1) and with vessels in the meninges that remained directly in electroblot. attached to the brain (data not shown). Reaction was con- Protein Electroblot Transfer to Nitrocellulose Paper. Sepa- fined to microvessels. This exclusive pattern of brain reac- rated proteins were transferred to nitrocellulose paper as tivity was retained in unfixed frozen, Bouins'-fixed Vibra- described by Towbin et al. (21). Blots that were to be stained tome sections, and paraffin-embedded sections from brains immunocytochemically were incubated in 3% (vol/vol) bo- that had been fixed in buffered 4% paraformaldehyde or vine serum albumin and 1% (vol/vol) normal sheep serum in fixatives containing 2% glutaraldehyde. At higher magnifi- 1.5 T buffer for 30 min while covered with aluminum foil to cation antibody binding appeared to be primarily associated avoid exposure to light. The blots were rinsed carefully for 15 with the luminal surface of the endothelial cell (Fig. 2) with min in 1.5 T buffer and cut into strips corresponding to only occasional faint reaction seen in other cell areas. separate lanes on the original gel. Immunocytochemical staining followed by histological coun- Immunocytochemistry. Strips of nitrocellulose paper were terstaining with hematoxylin/eosin (Fig. 3) confirmed that stained immunocytochemically by incubating with anti-EBA endothelia and not pericytes or basement membrane were ascites fluids, diluted 1:1000, for 1 hr; goat anti-mouse IgG, reacting. Microvessel reactivity was still seen at anti-EBA diluted 1:40, for 30 min; mouse peroxidase-antiperoxidase dilutions of 1:300,000 or greater, depending upon the ascites prepared from monoclonal antibody, diluted 1:200, for 30 fluid used. Additional sites of antibody binding examined min; and 0.05% diaminobenzidine tetrahydrochloride/0.01% were , and retina (Fig. 4). Endothelial hydrogen peroxide for 8 min. Each step was followed by optic nerve, careful rinsing for 15 min with several changes of 1.5 T buffer. PR *4 * $ . t

I ( * X w f a . . . 1. t ¢ k A

I - W # b; .4 In + 1 \ f ..v 4' A I N , 1 Ixt. .w,I

N , f

i I I

at . fI if I GC z ; *1 . r, a t FIG. 2. Brainstem microvessels. Anti-EBA reaction appeared to be predominantly associated with the luminal surface of the endo- FIG. 4. Reaction of anti-EBA with neural retina. Ganglion cell thelial cell (arrows). The occasional faint reaction with other areas of layer (GC) is at the bottom of the micrograph, and photoreceptor the cell is indicated by the arrowhead. (x480.) layer (PR) is at the top. (x230.) Downloaded by guest on September 27, 2021 Neurobiology: Stemberger and Stemberger Proc. NatL. Acad. Sci. USA 84 (1987) 8171

O

(.) K--- AP 'C-- " A

lb L 2 St

FIG. 5. Fenestrated blood vessels in the area postrema (AP), at the right in the micrograph, reacted poorly, or not at all with FIG. 7. At 3 days ofage, reaction was seen in some microvessels anti-EBA. Nonfenestrated vessels in surrounding brain reacted well. of the developing brain. These vessels are in the cortex. (x230.) Antibody dilution 1:1000. (x 120.) the luminal surface (Fig. 9). In the epidermis of the skin, anti-EBA reacted with small cells with fine processes (Fig. cells in the peripheral nervous system, where a blood-nerve 10). Reaction occurred at the cell surface and within the barrier is present, also reacted with anti-EBA. cytoplasm. These cells have tentatively been identified as Antibody binding to sites in the nervous system known to Langerhans cells, a special kind of macrophage, which are possess fenestrated blood vessels was absent, or detectable Ia-presenting cells. only at high concentrations ofantibody. In the area postrema, On nitrocellulose blots of electrophoresed preparations of one of the circumventricular organs, microvessels reacted brain microvessel membranes, the antibody identified three poorly, or not at all, even at a 1:1000 dilution of anti-EBA components with Mr 30,000, 25,000, and 23,500 (Fig. 11, (Fig. 5). Some vessels in the choroid plexus (Fig. 6), trigem- lanes C, D, and E). The lower Mr band was less easily inal ganglion, and pineal body were positive, but this reaction detected, and higher concentrations of the microvessel prep- was quickly abolished by dilution of the ascites fluid to aration were required to visualize this band. Anti-EBA did 1:10,000. not react with proteins in the endothelial cytosol fraction In the developing rat brain, sagittal sections from 0-, 3-, 6-, (Fig. 10, lanes A and B). No bands were seen on blots ofliver 11-, 15-, 20-, and 28-day rats were examined for appearance homogenates reacted with anti-EBA. of EBA. Antibody reaction with some microvessels was seen at 3 days (Fig. 7), and the number of positive vessels DISCUSSION increased with rapidly development. By 20 days, essentially Anti-EBA recognized endothelial proteins present predomi- all recognizable blood vessels reacted with anti-EBA. There nately in endothelia of the blood-brain and blood-nerve was no rostral/caudal gradient of first-appearing reactivity. barriers. Although some fenestrated endothelia, such as the Peripheral, non-nervous system tissues were examined capillaries of the choroid plexus, bound the antibody to a also. Blood vessels and sinusoids in the liver (Fig. 8) and limited extent, EBA appeared to be a minor component of vessels in the heart, adrenal, skeletal muscle, intestine, these vessels. The function and biochemical nature of EBA thymus, lymph nodes, pancreas, thyroid, skin, and pituitary are not yet known. It is possible that EBA proteins are were unreactive. In the spleen, a patchy reaction was seen on components of microvessel tight junctions. Pardridge et al. some vessel walls or associated with cells sometimes seen on (11) have suggested this function for a Mr 46,000 protein that had a ring-like pattern ofendothelial reactivity similar to that

i' jTAn

I 0 4A'01 N r) FIG. 6. Some endothelia of the choroid plexus were positive for anti-EBA reactivity at 1:5000 antibody dilution (arrow). Most choroid plexus endothelia were nonreactive at this dilution. Brain FIG. 8. Liver blood vessels and sinusoids did not react with microvessels were strongly stained. (x230.) anti-EBA. (x230.) Downloaded by guest on September 27, 2021 8172 Neurobiology: Sternberger and Sternberger Proc. Natl. Acad. Sci. USA 84 (1987)

4V

f 0~

,...... )

a1,f0 % t% % ! If .v9.' k I i A p .k

FIG. 9. In the spleen, antibody binding to portions of some blood vessels and some cells was seen (arrows). The diffuse background of non-anti-EBA-related staining (arrowhead), presumably due to Fc receptor binding, was also seen in control sections. (x 120.)

seen with anti-EBA. However, this pattern suggests that, at the light microscopic level, reaction was associated with the A B C D E entire luminal surface, rather than being restricted primarily to the lateral margins of contiguous endothelial cells. Alter- FIG. 11. Electroblot ofisolated microvessel cytosol (lanes A and natively, EBA may be associated with a receptor complex B) and membrane fraction (lanes C, D, and E). Lane A, 30 ,ug of characteristic of nervous system endothelia. cytosol fraction; lane B, 100 ,ug of cytosol fraction. Lanes C, D, and Anti-EBA was distinct from other monoclonal antibodies E were loaded with 40, 100, and 200 ,ug of membrane fraction. All that have been identified as reacting with brain endothelia lanes were treated with anti-EBA. (3-10). The unique patterns of reactivity on tissue sections and electroblots also eliminate its identification with known have apparent molecular weights similar to some populations of endothelial proteins such as factor VIII (22), transferrin Ia-like molecules (29). However, the presence of EBA on receptor (23), and other proteins that are characteristic of all normal brain endothelia and the absence of EBA on vessels in endothelial cells and have different molecular weights. experimental allergic encephalomyelitis that are surrounded by It is intriguing that anti-EBA was expressed on some cell inflammatory cell infiltrations (30) make it highly unlikely that types and tissues associated with the immune system. En- the antibody is recognizing Ia antigens. Instead, these findings dothelial cells are considered to play a major role in the suggest that anti-EBA recognized an endothelial cell modifica- development of cell-mediated immune responses (24). After tion in inflammatory disease and that in vivo modification of inducible expression of class II major histocompatibility EBA may be part of the local immune response. antigens (Ia), endothelia can act as antigen-presenting cells At birth, the rat has poor vascularization of the brain. (25-27). Langerhans cells, in the epidermis of the skin, also Vascularization proceeds by a budding or sprouting from a role in contact existing vessels (31). There is evidence that a selective barrier function in Ia presentation and play key for some molecules exists in the newborn rat and that this sensitivity reactions in the skin (28). It was not clear from our barrier increases in efficiency with age. The postnatal ap- preliminary observations on spleen sections what immune pearance of EBA may reflect the gradual biochemical devel- cell types were recognized by anti-EBA. opment of permeability barriers that accompany the vas- The bands reacting with anti-EBA on nitrocellulose blots cularization of brain. A complete vascular bed was estab- lished by 3 weeks (31), approximately the time when the adult pattern of anti-EBA reactivity is seen. Many ofthe proteins expressed by brain endothelia have not yet been characterized. Our monoclonal antibody recognized plasma membrane components not expressed by non-nervous N.. system endothelia. Anti-EBA will provide a means of separat- Ik ing and characterizing unique endothelial cell proteins that may (i) play a role in the biochemical definition of the blood-brain barrier and (ii) participate in pathological alterations of the barrier that occur in cell-mediated responses. We thank Roberta Shinaberry for her excellent assistance in preparing the microvessel fractions. This research was supported by grants from the National Multiple Sclerosis Society, the National Institutes of Health (NS 24185 and NS 21681), and the National Science Foundation (BNS 85-06474). 1. Sternberger, L. A. & Sternberger, N. H. (1983) Proc. Natl. Acad. Sci. USA 80, 6126-6130. FIG. 10. Small cells with fine processes (arrow) were stained by 2. Brightman, M. W. (1977) Exp. Eye Res. Suppl. 25, 1-25. anti-EBA in sections of rat skin. These cells may be Langerhans 3. Michalak, T., White, F. P., Gard, A. L. & Dutton, G. R. cells. Skin blood vessels did not react. (x480.) (1986) Brain Res. 379, 320-328. Downloaded by guest on September 27, 2021 Neurobiology: Stemberger and Sternberger Proc. Natl. Acad. Sci. USA 84 (1987) 8173 4. Werner, R., Hallmann, R., Albrecht, U. & Henke-Fahle, S. 18. Smith, P. K., Krohn, R. I., Hermanson, G. T., Mallia, A. K., (1986) EMBO J. 5, 3179-3183. Gartner, F. H., Provenzano, M. D., Fujimoto, N. M., Olson, 5. Hamburger, A. W., Reid, Y. A., Pelle, B. A., Breth, L. A., B. J. & Klenk, D. C. (1985) Analyt. Biochem. 150, 76-85. Beg, N., Ryan, U. & Cines, D. B. (1985) Tissue Cell 17 (4), 19. Laemmli, U. K. (1970) Nature (London) 227, 680-685. 451-459. 20. Goldstein, M. E., Stemberger, L. A. & Sternberger, N. H. 6. Alles, J. U. & Bosslet, K. (1986) J. Histochem. Cytochem. 34 (1983) Proc. Natl. Acad. Sci. USA 80, 3101-3105. (2), 209-214. 21. Towbin, H., Staehelin, T. & Gordon, J. (1979) Proc. Natl. 7. Schlingemann, R. O., Dingjan, G. M., Emeis, J. J., Blok, J., Acad. Sci. USA 76, 4350-4354. Warnaar, S. 0. & Ruiter, D. J. (1985) Lab. Invest. 52 (1), 22. Mukai, K., Rosai, J. & Burgdorf, W. H. C. (1980) Am. J. 71-76. Surg. Pathol. 4, 273-279. 8. Arvieux, J., Willis, A. C. & Williams, A. F. (1986) Mol. 23. Jefferies, W. A., Brandon, M. R., Hunt, S. V., Williams, Immunol. 23 (9), 983-990. A. F., Gatter, K. C. & Mason, D. Y. (1984) Nature (London) 9. Ghandour, S., Langley, K., Gombos, G., Him, M., Hirsch, 312, 162-163. M.-R. & Goridis, C. (1982) J. Histochem. Cytochem. 30 (2), 24. Burger, D. R. & Vetto, R. M. (1982) Cell. (London) Immunol. 165-170. 70, 357-361. 10. Labistie, M.-C., Poole, T. J., Peault, B. M. & Le Douarin, 25. Wagner, C. R., Vetto, R. M. & Burger, D. R. (1984) Immuno- N. M. (1986) Proc. Natl. Acad. Sci. USA 83, 9016-9020. biology 168, 453-469. 11. Pardridge, W. M., Yang, J., Eisenberg, J. & Mietus, L. J. 26. Pober, J. S., Collins, T., Gimbrone, M. A., Jr., Libby, P. & (1986) J. Cereb. Blood Flow Metab. 6, 203-211. Reiss, C. S. (1986) Transplantation 41 (2), 141-146. 12. Stemberger, L. A., Harwell, L. W. & Stemberger, N. H. 27. Sobel, R. A., Blanchette, B. W., Bhan, A. K. & Colvin, R. B. (1982) Proc. Natl. Acad. Sci. USA 79, 1326-1330. (1984) J. Immunol. 132, 2402-2411. 13. Kohler, G. & Milstein, C. (1976) Eur. J. Immunol. 6, 511-519. 28. Silberberg, I., Baer, R. L. & Rosenthal, S. A. (1976) J. Invest. 14. Sternberger, L. A., Hardy, P. H., Cuculis, J. J. & Meyer, Dermatol. 66, 210-235. H. G. (1986) J. Histochem. Cytochem. 18, 315-333. 29. Lampson, L. A. & Levy, R. (1980) J. Immunol. 125 (1), 15. Sternberger, L. A. (1986) Immunocytochemistry (Wiley, New 293-299. York), 3rd Ed. 30. Sternberger, N. H., Murant, F. G., Parkison, J. A., Kies, M. 16. Mrsulja, B. B., Mrsulja, B. J., Fujimoto, T., Klatzo, I. & & Sternberger, L. A. (1984) J. Neuropathol. Exp. Neurol. 41, Spatz, M. (1976) Brain Res. 110, 361-365. 351 (abstr.). 17. Lidinsky, W. A. & Drewes, L. R. (1983) J. Neurochem. 41, 31. Nousek-Goebl, N. A. & Press, M. F. (1984) Dev. Brain Res. 1341-1348. 30, 67-73. Downloaded by guest on September 27, 2021